Why you forget last week's questions, and how spaced review brings them back
The research on spacing and testing, and how Pradis uses three stars to decide when each question returns.
On Monday you work through a set of biostatistics questions and finally get it: lower the fasting glucose cut-off for diabetes from 126 to 100 mg/dL, and sensitivity and negative predictive value go up while specificity and positive predictive value go down. Five weeks later, on exam day, you can't remember which way it goes, only that you once knew. That isn't a willpower problem. It is how memory works, and it has been measured for well over a century.
The forgetting curve
In 1885 the German psychologist Hermann Ebbinghaus memorized lists of nonsense syllables and tested himself at intervals from twenty minutes to a month. Most of what he lost, he lost early: a large share within the first day, then more slowly after that. The shape he found, a steep drop that flattens out, has held up in many studies since.
The useful part is what happens when you recall something before it's gone. Each successful recall makes the next drop slower. So the aim is to bring a fact back just as you're about to lose it, and to wait a little longer each time.
Two effects that do the work
Spacing
The same amount of study spread over several days beats the same amount crammed into one sitting, for anything you need to remember longer than a few days. A 2006 review by Cepeda and colleagues pooled more than 300 experiments and found the advantage again and again. Their later work showed that the best gap grows with how long you need to remember. If the exam is months away, reviews should be weeks apart, not hours.
Testing
Pulling an answer out of memory strengthens it more than reading it again. In a well-known 2006 study, Roediger and Karpicke had students either reread a passage or practice recalling it. Rereading looked better on a test a few minutes later. A week later, the students who had practiced recalling remembered more.
Medicine has its own evidence. In a randomized trial published in Medical Education in 2007, Kerfoot and colleagues found that short clinical questions sent to medical students by email, repeated over weeks, improved how much they retained.
Why a question bank alone doesn't do this
Most question banks show you a question once. You answer, read the explanation and move on. The next time you meet that question may be never, or a re-do of the whole block weeks later, when you half remember the answer letter instead of the reasoning. Neither one is spacing, and recognizing a familiar stem is different from recalling the answer.
Understand first, then come back
Spacing makes memories last, but it can't make them make sense. That part comes first. Pradis is built on A-WIN, the Thieme Step 2 CK review by Manoj Gurung, MD and Yayra Musabek, MD, academic hospitalists. The book puts one principle plainly: don't build buildings without foundations, and avoid rote memorization; use mnemonics, associative memory and memory hooks as much as possible.
So most Pradis questions are tables of cases that look alike, and every row has its own "Why": the clue that decides that case and not its neighbor. One Biostatistics table, for example, shows three cut-offs for HbA1c and asks, row by row, which gives the most false negatives, which the most false positives, and which the highest sensitivity. The "Why" for the last row: the lowest cut-off calls the most people positive, so it misses the fewest people with diabetes. Memory hooks sit beside the facts that need them, like "SN-N and SP-P" (sensitivity moves with negative predictive value, specificity with positive predictive value), "DAN does cocaine" for the treatment of cocaine chest pain, or "what C-peptide is to insulin, thyroglobulin is to thyroid hormone". Spaced review is the second half. Once a distinction makes sense, it brings the question back at the moment you are about to forget it.
How Pradis decides when a question comes back
Every question in Pradis earns up to three stars. A star needs a correct answer you were sure of, given on a day the question is due.
| What happens | Stars | The question comes back |
|---|---|---|
| First answer, correct and sure | ★☆☆ | in 5 days |
| Correct and sure again | ★★☆ | in about 3 weeks |
| Correct and sure a third time | ★★★ mastered | in about 4 weeks, then about every 8 weeks |
| Wrong, at any point | back to ☆☆☆ | tomorrow |
| Correct, but you ticked "I wasn't really sure" | no new star | in 2 days |
On the standard schedule the three stars land on day 0, day 5 and about day 26. That is what "mastered" means in Pradis: right on three different days, without guessing, with growing gaps in between.
A few rules keep the stars honest:
- Only scheduled reviews earn stars. Practicing a question again before it's due is fine, but it doesn't add a star. You can't earn three stars in one evening.
- A wrong answer always counts. Miss it early or late, and the question starts over and comes back tomorrow.
- "I wasn't really sure" costs nothing. You still see the explanation. A lucky guess isn't memory, so the question comes back in 2 days instead of earning a star.
- Each table row is its own question. In a table of look-alike diseases, the row you keep confusing comes back on its own schedule while the rows you know move on. The free Biostatistics chapter alone has 24 tables with 88 rows, and each row earns its own stars.
- Mastered questions can slip. Miss a mastered question and it's marked "Needs refresh". Answer it correctly the next day and again 3 days later, and it's mastered again.
Planned around your exam date
Spacing assumes you have time, and before an exam you may not. When you set your exam date, Pradis shortens the gaps so no review lands after the exam: no question waits longer than half of the days you have left. In the weeks before the exam, a final check brings every question back once, weakest first, and the last few days are kept for repairing what the final check finds.
Your review list shows what is due today, up to your daily limit (30 to start; you can set anything from 5 to 50). If you've been away, Pradis spreads the backlog over the next few days instead of handing you hundreds at once. You can also pause reviews; when you come back, every due date moves forward by the length of the pause.
What we leave out on purpose
No streaks, no points and no predicted score. Mastered counts only questions you got right, without guessing, at three spaced reviews. Questions that slipped are shown separately as "Needs refresh", so a mastered count never includes a question you've since missed.
How to get the most from it
- Do today's reviews before new questions. Reviews are the questions you're about to forget.
- Tick "I wasn't really sure" whenever you guessed. It only helps you.
- Set your exam date in Study settings, so the schedule fits before it.
- After a miss, read the row's "Why", not just the right answer. The clue that separates the look-alikes is what the next review tests.
The Biostatistics chapter is free once you log in, with spaced review switched on: 122 questions across 18 topics, from sensitivity and specificity to biases. Answer a few today, and Pradis will tell you when to see them again.
References: Ebbinghaus H. Memory: A Contribution to Experimental Psychology. 1885. Cepeda NJ, et al. Distributed practice in verbal recall tasks: a review and quantitative synthesis. Psychol Bull. 2006;132(3):354-380. Cepeda NJ, et al. Spacing effects in learning: a temporal ridgeline of optimal retention. Psychol Sci. 2008;19(11):1095-1102. Roediger HL, Karpicke JD. Test-enhanced learning: taking memory tests improves long-term retention. Psychol Sci. 2006;17(3):249-255. Kerfoot BP, et al. Spaced education improves the retention of clinical knowledge by medical students: a randomised controlled trial. Med Educ. 2007;41(1):23-31.